Combining Simultaneously Measured UV and IR Radiances from OMI and TES to

Improve Tropospheric Ozone Profile Retrievals

Dejian Fu1, John Worden1, Susan Kulawik1, Kevin Bowman1, Stanley P. Sander1

1Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA

Xiong Liu2

2Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA

Introduction

Worden J., X. Liu, K. Bowman, K. Chance, R. Beer, A. Eldering, M. Gunson, and H. Worden, Improved tropospheric ozone profile retrievals using OMI and TES radiances, GRL, 34, 2007.

Landgraf, J., and O. P. Hasekamp, Retrieval of tropospheric ozone: The synergistic use of thermal infrared emission and ultraviolet reflectivity measurements from space, JGR, 112, 2007.

Data Source Vertical Resolution

GEOS – CHEM True profile

TIR + UV 3.6 – 4.8 km

TIR 6 km

UV 12 km

X_ret = Xa + A (X_true - Xa)

Tropospheric Emission Spectrometry and Ozone Monitoring Instrument

TES is a Fourier transform spectrometer Spectral Range: 650 – 3050 cm-1 (or 15.38 to 3.28 micron) Spectral Resolution: 0.06 cm-1

Ground Pixel Size: 8.3 by 5.3 km2 (along track by cross track)

OMI is an imaging grating spectrometer Spectral Range 270 – 500 nm (37,037 – 20,000 cm-1)

UV1 module (264 – 311 nm): 30 pixels UV2 module (307 – 383 nm): 60 pixels Visible module (349 – 504 nm): 60 pixels

Spectral Resolution: 0.42–0.63 nm Ground Pixel Size:

UV1: 13 by 48 km2 (along track by cross track) UV2: 13 by 24 km2 (along track by cross track) Visible: 13 by 24 km2 (along track by cross track)

OMI and TES Measurements Over Houston, Texas On May 9th, 2009

23.1 km

16.4 km11.2 km9.5 km

Criteria of Coincident Measurements Cloud optical depth: <= 0.005 Ozone sonde measurements are available for comparisons Time difference of OMI and TES measurements to ozone sonde: 0.5 hours Distances of OMI and TES measurements to ozone sonde site: 25 km

Spectral Regions Used in Joint Retrievals

TES

OMIUsed in this work

TES and OMI Normalized Radiances Measured Over Houston, on May 9th, 2009

ConvergenceYes

Output Retrieved Parameters [O3]

Update [O3] and Other Parameters

to be Retrieved

OMI FM Model

No

UV Simulated Radiances

TIR Measured Radiances UV Measured Radiances

TIR Simulated Radiances

Call Forward

Model

Retrievals: From Measured Spectra to O3 Profiles

TES FM Model

Retrievals: From Measured Spectra to O3 Profiles

Radiative Transfer Models OMI

Uses the linearized vector radiative transfer model VLIDORT [Spurr, 2006] for the numerical computation of the Stokes vector in a multiply-scattering multilayer medium

TES Simulates Earth radiances in the TIR spectral region, has been

developed and validated by the TES retrieval team

Retrieval Methodology Follows the optimal estimation method [Rodgers 2000]. Levenberg-Marquardt algorithm [Rodgers 2000; Bowman et al, 2006] was

applied to the combined TES/OMI radiances

Instrument

Fitting VariablesNumber

VariablesA Priori

A Priori Error

TES + OMI O3 at each layer 25 MOZART-3 MOZART-3

TES

H2O at each layer 16 GEOS4 NCEP ~30%

Surface temperature 1 GEOS4 0.5K

Surface emissivity 32

ASTER and

landuse map

~0.006

Cloud extinction 10Initial BT difference

300%

Cloud top pressure 1 500 mbar 100%

OMI

UV1 Surface Albedo 1 climatology 0.05

UV2 Surface Albedo 2 climatology 0.05

first-order wavelength-dependent term for UV2 Surface Albedo

1 0.0 0.01

Ring scaling parameters 2 1.9 1.0

Radiance/irradiance wavelength shifts 2 0.0 0.02 nm

Radiance/O3 cross section wavelength shifts (zero order)

2 0.0 0.02 nm

Radiance/O3 cross section wavelength shifts (first order)

2 1.0 0.004

Radiance Residuals (Measured – Modeled) /Noise Estimate

First Iteration RMS: 6.52Last Iteration RMS: 1.33

First Iteration

Last Iteration

Vertical Sensitivity and Resolution of O3 Observations

Vertical Sensitivity and Resolution of O3 Observations

Temperature Profile over Houston, Texas, USA on May 9th, 2009

High Surface Temperature High Thermal Contrast

O3 Profiles Over Houston On May 9th, 2009

8:35 AM

Error Analysis and Characterization

Summary and Future Work

Summary This work demonstrates that ozone profiles can be estimated from

combined OMI and TES radiances Combined OMI/TES retrieval converges with chi 1.3; and ozone profile

consistent with co-located sonde measurement Vertical resolution increases: averaging kernels are more separated in

altitude for combined retrieval Sensitivity (DOFS) to tropospheric variability improves from 1.76 to 1.92

Future work To further improve the program package in order to do large volume of

data processing Compare PBL ozone estimates using combined OMI/TES retrieval to in

situ measurements for spring / fall conditions (lower thermal contrast / higher albedo). Under these conditions we expect larger improvements in vertical resolution for combined retrieval.

Acknowledgements

Drs. Quitus L. Kleipool and Maarten Sneep at Royal Netherlands Meteorological Institute (KNMI)

Drs. Alyn Lambert and Robert Herman, Vijay Natraj at JPL

Your Attention

Program Package

UV RTM

ConvergenceYes

Output Retrieved Parameters [O3]

Update [O3] and Other Parameters

to be Retrieved

OMI Instrument

Model

No

UV Simulated Radiances

TIR Measured Radiances UV-VIS Measured Radiances

OMI L1B Data Products

TIR Simulated Radiances

Call Forward

Model

TIR RTMAtmospheric State P, T, [O3]

Viewing Geometry

Surface Albedo

Retrievals: From Measured Spectra to O3 Profiles

OMI L1B Data Products

TES L1B Data Products

Viewing Geometry

Emissivity &Surface T,

H2O

TES Instrument

Model

Program Package

TES L1B Data Products

Retrievals Using TES and OMI Measurements over Egbert Canada on July 5th, 2008

35.7 km

12.1 km22.0 km

3.0 km

Criteria of Coincident Measurements Cloud optical depth: <= 0.005 Ozone sonde measurements are available for comparisons Time difference of OMI and TES measurements to ozone sonde: 3 hours Distances of OMI and TES measurements to ozone sonde site: 50 km Surface albedo was near zero over water surface